Sep 12, 2020

Huge Idea from Yuval Hariri

 

Another hugely important idea from one of our greatest contemporary intellectuals. We may find ourselves nostalgic for the comparatively simple time when wealth and resources were the primary distinction between classes.

Despite being a longtime proponent of transhumanist technologies, it's impossible to ignore the enormous risk and friction they'll likely impose on society:



Apr 8, 2020

Are We Too Busy Dying from Coronavirus to Die From Anything Else?

Estimates of up to 2 million deaths in the United States, mortality rates trending at 3%, and new infections growing rapidly into March were enough to bring much of the country to a grinding halt. 

Kinsa's new Health Weather Map
Recent predictions however are trending dramatically lower, with 70k-100k deaths anticipated this year. This could be evidence of the power in social distancing, but could also speak to a much less lethal disease than originally feared. If there’s any merit to the idea I’m about to share, then we’ll see mortality rates driven dramatically lower still.

Until we better understand the true number of infected, we won’t a know the actual infected fatality rate (IFR) for coronavirus; but with an estimated 85% of coronavirus-positive patients believed to suffer mild (or no symptoms at all), it’s possible there’s an already huge number of overall infected coronavirus patient pool that’ll radically reduce the current trending mortality rate of ~3%.

There’s another factor, however, one that could lower mortality rates even further, and as we move into April, it’s receiving effectively zero press attention. It has to do with how we’re classifying coronavirus fatalities in patients with already deadly diseases (“comorbidities” such as advanced cardiovascular and respiratory disease).  

Given roughly 94% of coronavirus deaths are in patients with an already potentially deadly disease [UPDATED LINK], how we report underlying causes of death in coronavirus-positive patients is hugely consequential and, as presently organized, is hugely flawed.

The United States relies on the “NVSS” (National Vital Statistics System) to set standards for the designation and reporting of underlying cause of death nationwide. Such data is aggregated and, together, forms the basis for determining mortality rates for the things that kill us. The NVSS is managed by the CDC, and serves as a model for similar agencies and reporting systems around the world.

In the context of coronavirus fatality reporting, their standards may be contributing a unique and potentially enormous vulnerability in the way we consider coronavirus’ potential for harm. It all has to do with a fairly inane seeming detail around how we attribute fatality reporting: by that I mean, we pair each death with a single, primary underlying cause.

Causes of death USA 2017
Said another way, our national fatality reporting system has a single-attribution bias. When we die, a coroner or medical examiner will list one, single underlying cause of death as the cause; even in instances where comorbidities played a strong (even dominant) contributory role in killing us. 

By it's very nature, a single-cause biased system cannot account adequately for the contributions coexisting diseases play. But for all its flaws, the system worked well-enough in peacetime. Today however, the system seems wildly unfit to help us understand a virus whose very lethality all but requires deadly preexisting comorbidities be present.

The coronavirus data being acted on is noisy, the stakes are high, so radical action may ultimately be justifiable. However, the relative softness of certain key variables (like the number of actual infections vs test-confirmed cases) alone are enough to haunt policymakers. Nobody paying attention would be surprised to see mortality rates drop dramatically owing to that factor alone. Now layer-in the structural flaw in how we’re counting coronavirus fatalities at the most fundamental level, it’s not hard to imagine coronavirus approaching the proverbial “n word” of comparisons: “flu like mortality”

What might that look like on paper? If it turns out the number of infected are exponentially higher than confirmed cases (which isn’t hard to imagine given coronavirus manifests as with light or no symptoms in 85% of those infected), and if we controlled for the single cause of death attribution bias, we could see the mortality rate go from 3% to as low as .12%.



The bias is real. The overwhelming super-majority of fatalities from coronavirus are in patients with preexisting, often deadly underlying conditions.

I’m suggesting we stay vigilant to the possibility we're so busy dying from coronavirus we forget to notice we’re not dying from everything else.


April 7, 2018




Supplemental Thoughts:

As of last week, the National Vital Statistics System issued guidance that instructed Medical Examiners and Coroners nationwide to identify Coronavirus disease deaths using the ICD–10 code “U07.1”. Specifically, “Deaths are coded to U07.1 when coronavirus disease 2019 or COVID-19 are reported as a cause that contributed to death on the death certificate. These can include laboratory confirmed cases, as well as cases without laboratory confirmation. If the certifier suspects COVID-19 or determines it was likely (e.g., the circumstances were compelling within a reasonable degree of certainty), they can report COVID-19 as “probable” or “presumed” on the death certificate.”

That means deaths from coronavirus needn’t be confirmed with testing to be counted.

For context, roughly 2.8mm people die in the United States every year (about 65mm globally). Each death is assigned a primary underlying cause of death and made available to various stakeholders, from government to media.

My concern has to do with our single-attribution-minded model for classifying underlying cause of death in coronavirus fatalities. Specifically, I haven’t been able to determine if medical examiners and coroners are controlling for possible bias in attributing fatalities to coronavirus when there are one or more comorbidities present (as is the case in ~94% of coronavirus fatalities). 

Heart disease is the leading cause of death in the United States. Every year it claims roughly 650k Americans. The fourth leading cause (behind cancer and accidents) are chronic respiratory diseases. They kill 160k Americans. Coronavirus requires the presence of those deadly diseases (or similarly deadly comorbidities such as cerebrovascular disease, diabetes, and pneumonia), to kill 95% of the time. It’s well established that coronavirus exacerbates both chronic respiratory and underlying cardiovascular illness, and while the data isn’t yet in, we’ll almost certainly see a significant drop in fatalities from the typical leading causes of death. How big a drop is the $20 trillion dollar question.

The New York Times recently wrote a story entitled “Where Have All the Heart Attacks Gone?”.  

While ~85% of heart attack patients typically go home, 10-15% don’t (roughly 647,000 per year). It’s a grim reality that coronavirus patients predisposed to heart attacks are more likely to suffer a fatal heart attack. 

Similarly, stroke specialists in New Orleans, Chicago, Seattle, report seeing stroke and transient ischemic attack (TIA) patient counts down 30% to as much as 50%. Fear of exposure to coronavirus at hospital, could be a contributing factor, but doesn’t explain a similar reduction in patients being experienced in telestroke and virtual care networks.

It seems a distinct possibility that coronavirus may hastening death in a significant number (or even a majority) of patients who would otherwise die this year.



Supplemental Resources and Reporting Tools:

Here are some of the data I regularly consult, as well some interesting articles and useful reporting technology.

Primary sources of data for new infections/fatalities:



New reporting technology that uses anonymized data from cell phones to score actual compliance with social distancing by area:



Unacast Social Distancing Scoreboard
Two articles: the first concerns likelihood that fatality rates are exaggerated by an order of magnitude. The other is a revised estimate from the same Imperial College scientist who scared the s*** out of the world a week ago with estimates of millions dead (now down to 20k or a few thousand in the UK). His last estimate won the Imperial College a $10mm grant from UK gov't.



At minute 38:40 in this podcast Marshall Burke, an assistant professor at Stanford's Department of Earth System Science, says the better air quality caused by shutdown in China will likely save 50,000 lives in China”:

(It’s worth noting those are just the impacts from temporary suspension of pollution activity.  They exclude reduction in traffic fatalities, reduction in other communicable disease fatalities from isolation measures, etc.) The main takeaway being this: running our world, as it is today, costs a lot of lives.

Here's a similar article:

And here are assumptions and a forecast on the pandemic from HealthData (interestingly, I noticed they revised their death forecast down..I have screenshot one forecast pre-revision...no law says you can't revise, but seems a bit shady to not account publicly for errors)


And lastly, here's a very cool utility that uses millions of anonymized IoT thermometer readings to create a national "health weather map":





Christian Hunter's Twitter Latest